Sora Background Information:

Sora, an AI video generation model released by the American artificial intelligence research company OpenAI, was officially launched to the public on February 15, 2024 (local time in the U.S.). OpenAI does not merely regard it as a video model but as a “world simulator”.

The name Sora is derived from the Japanese word “空” (sora), meaning sky, symbolizing its infinite creative potential. The technology behind it was developed based on OpenAI’s text-to-image generation model DALL-E.

Sora can create realistic videos of up to 60 seconds based on user text prompts. The model understands how these objects exist in the physical world, enabling it to deeply simulate real-world physics and generate complex scenes with multiple characters and specific movements. It inherits the image quality and instruction-following capabilities of DALL-E 3, allowing it to comprehend the requirements posed by users in prompts.

Sora brings limitless possibilities for artists, filmmakers, or students needing to produce videos. It is a step in OpenAI’s plan to “teach AI to understand and simulate the physical world in motion” and marks a leap in AI’s ability to understand and interact with real-world scenes.

Success of OpenAI in Large Models

At the end of 2022, OpenAI officially launched ChatGPT, a natural language processing tool powered by AI technology that can engage in conversation by learning and understanding human language. ChatGPT was the first step taken by OpenAI, showcasing the potential of AI to everyone through this phenomenal product, which demonstrated a leap in understanding and logical capabilities compared to past AI.

Advancements in Visual Algorithms

Recent breakthroughs in visual algorithms have made progress in generalization, promptability, generation quality, and stability, indicating the approach of a technological turning point and the emergence of blockbuster applications. Particularly in 3D asset generation and video generation, these fields have greatly benefited from the maturity of diffusion algorithms. However, compared to image generation, 3D asset and video generation face more challenges in terms of data and algorithms.

Nonetheless, considering the accelerating role of large language models (LLMs) in various AI fields and the emergence of excellent open-source models, the industry is expected to achieve greater development in 2024. From late 2023 to early 2024, AI-generated video applications like Pika and HeyGen have gradually gained attention, validating the continuous progress and maturity of multimodal technology. However, at the same time, advocates for democracy and AI researchers warn that these tools have already been used to deceive and mislead the public.

Sora is Here: Who Will Lose Their Jobs?

Screenshot from OpenAI’s official website

By inputting just a few words, you can generate stunning 60-second videos! The AI video generation model Sora released by the American OpenAI has recently shocked the world.

Just over a year after OpenAI launched ChatGPT, it has thrown another bombshell. What is so amazing about the “text-to-video” AI model Sora? What impact does it have on the industry? What hidden dangers exist? Xinhua News Agency reporters provide insights for you —

What Can Sora Do?

The multiple short videos generated by Sora have gone viral online, featuring realistic and smooth scenes with rich details.

This is OpenAI’s first foray into the AI video generation field. According to the company, Sora uses a Transformer architecture to create realistic and imaginative scenes based on text instructions, generating various styles and different aspect ratios, with videos lasting up to one minute.

In other words, give Sora some prompts, and it can produce a high-quality short video. Let’s experience Sora’s creative power together.

Images of videos generated by Sora released on OpenAI’s official website show an SUV driving on a winding mountain road.

Sora can also generate videos based on static images, extend existing videos, or fill in missing frames.

OpenAI states that Sora can deeply understand language, comprehending not only user text prompts but also how the mentioned objects exist in the physical world. “We are teaching AI to understand and simulate the physics of motion, aiming to train models to help people solve problems that require interaction with the real world.”

However, are the video works showcased on OpenAI’s website the average or the highest standard of Sora’s creations?

OpenAI admits that the videos generated by Sora may currently contain illogical images, confuse left and right spatial details, and struggle to accurately simulate the physical principles and causal relationships in complex scenes. For example, a person may take a bite of a cookie, but there are no bite marks on the cookie. Nevertheless, with enhanced computing power and model improvements, people may soon obtain more refined and advanced video generation capabilities.

Who Might Lose Their Jobs?

OpenAI’s launch of Sora seems more like a preview, and the public still finds it difficult to gain a comprehensive understanding of the model’s strengths and weaknesses. OpenAI has stated that currently, Sora is mainly accessible to specific groups such as designers and filmmakers to gather feedback for improving the model. The company has not disclosed the data used to train the Sora model or any basic details, nor has it set a timeline for public release.

Sora Video Generation Screenshot

Some analysts believe that Sora again highlights the profound impact of advancements in AI technology on real life and traditional industries. The enormous growth potential of AI in the video generation field not only opens the door to shaping new business models in the film and television industry but may also disrupt the existing film and television industry.

The day after Sora’s launch, the stock price of Adobe, a company specializing in image processing and video production software, fell over 7%.

Hollywood faced its first industry-wide strike by writers and actors in 63 years last year, as some job opportunities in the industry may be replaced by AI. The emergence of Sora has made this threat feel more immediate and real.

Making Forgery More Realistic and Hard to Distinguish

OpenAI has described the video generation model in its technical report as a “world simulator”.

If the world can be simulated, where is the boundary between true and false? Many industry insiders worry that Sora will fuel the rise of “deepfake” technology. Farid, the Deputy Dean of the Information School at the University of California, Berkeley, stated: “When news, images, audio, and video — anything can be forged, then in that world, nothing is real.”

Image taken on November 2, 2023, at the first AI Safety Summit in Bletchley Park, UK, showing a participant passing by the promotional display board. Xinhua News Agency reporter Li Ying reported

In response to concerns about forgery, OpenAI stated that when it officially releases the product to the public, it will ensure that generated videos include source metadata and will launch tools to detect the authenticity of videos. OpenAI also promised to implement safety measures before using Sora in products, including adversarial testing of the model by experts in fields such as misinformation, hate content, and bias to assess potential harm or risks; verifying and rejecting text input prompts that include extreme violence, sexual content, hate images, and other IP.

However, OpenAI acknowledges that even with extensive research and testing, “we cannot predict all the beneficial ways people will use our technology and all the ways they may misuse it.”

Can AI’s Rapid Growth Be Controlled?

Disruptive innovations continue to emerge in the technology sector, and how to achieve a balance between embracing technological progress and ensuring social safety is increasingly attracting attention from all walks of life.

OpenAI stated that it will collaborate with policymakers, educators, and artists worldwide to understand their concerns, identify positive use cases for Sora, and believes that learning from real-world usage is a key component in creating and releasing increasingly safe AI systems.

On July 7, 2023, at the “AI for Humanity Global Summit” in Geneva, a participant interacted with the simulation robot Sophia. Xinhua News Agency reporter Lian Yi reported

Industry insiders point out that in the current governance framework, where control measures have not kept pace, relying solely on companies may not provide the safety and trustworthiness of AI that society needs.

Sora’s Popularity! How Will AI Impact the World?

Clear and lively eyes, adorable pets, the mysterious underwater world, bustling summer streets, and a futuristic magical city…

This short video, featuring realistic scenes, rich colors, and a strong atmosphere, was entirely created by an AI system.

Recently, the American OpenAI released the first video generation model “Sora”. This model can generate a 60-second short video by receiving text instructions. A year ago, the same research center released the AI language model ChatGPT, which made writing and creating text, checking code programs, etc., as easy as pie.

AI chat, AI painting, AI music… With the emergence of a series of AIGC (AI-generated content), the so-called “AI revolution” that has a disruptive impact on modern social life has officially arrived.

What abilities does AI really have? Why does each iteration and upgrade spark global discussions?

Generative AI can transform input content

into novels,movies, and artworks.

Google’s AI model“Bard” can quickly generate a short story or poem based on multiple words you input.

In February of this year, Google announced that “Bard” has been renamed “Gemini.” This is a multimodal large model that can understand and combine different types of information, including text, code, audio, images, and video.

“Bard” read almost all content on the internet in a few months and developed a large language model, providing answers based on the language model rather than web searches.

DALL-Ecan turn any content you input into an artwork.

To train DALL-E, the developing company provided it with about 600 million labeled images from the internet. Through deep learning, it can not only understand individual objects but also learn the relationships between different objects.

Using Runway, you can generate visual effects that typically take days to complete in just seconds.

The company’s founder, Barrensuela, frankly stated that with the support of generative AI, the thresholds and costs of film production will be greatly reduced in the future.

Biomedical, autonomous driving, weather forecasting…

The AI technology market is vast.

In addition to the field of artistic creation, AI technology is also worth attention for its applications in medicine, urban services, and weather forecasting.

On January 29, American entrepreneur Musk stated that his brain-computer interface company “Neuralink” has completed the first human transplant of a brain-computer interface device, and the transplant recipient is in good condition. It is reported that this technology is fully implantable, battery-powered, and wireless, connecting via Bluetooth throughout.

On January 30, Tsinghua University announced that its brain-computer interface research team, in collaboration with Xuanwu Hospital of Capital Medical University, successfully conducted the world’s first clinical trial of a wireless minimally invasive brain-computer interface in October 2023.

A patient who had been paralyzed for 14 years due to spinal cord injury after a car accident has achieved brain-controlled functions such as drinking water independently after three months of rehabilitation training, with a gripping accuracy of over 90%.

Although brain-computer interface technology still faces many challenges and even skepticism, it is undeniable that artificial intelligence has achieved remarkable successes in the medical field, especially in diagnostics based on medical imaging. Currently, the U.S. Food and Drug Administration has approved about 420 algorithms related to imaging, mainly used for cancer treatment, with an accuracy rate of 80% to 90%.

Besides the medical field, generative AI will also participate more broadly in urban public services and weather forecasting practices.

Kumar is a truck driver in India. When driving on the highway, a round trip takes 60 hours, and long periods of fatigue driving can easily lead to traffic accidents. Now, he has a “silent” companion on his work journey, reminding him to avoid fatigue or pay attention to distance.

This is a terminal device powered by AI and computer vision technology. The outward-facing camera can measure the distance between the vehicle and other objects. The inward-facing camera monitors the driver’s behavior and condition; if the driver is on the phone or drowsy, the device will alert the driver to drive properly.

In July 2023, Huawei Cloud’s Pangu meteorological large model officially launched on the European medium-term weather forecasting official website, showcasing China’s large model’s ability to solve problems in the meteorological field.

The World Bank estimates that improving weather forecasting and early warning systems can bring an annual value of $162 billion and save about 23,000 lives.

Furthermore, AI technology is playing an increasingly important role in promoting educational equity and addressing the challenges of an aging society, forming a huge market scale. According to Bloomberg, the market scale of generative AI is expected to expand to $1.3 trillion by 2032.

AI-generated deceptive content interferes with elections

or creates chaos during national elections.

While AI technology brings many new opportunities, it inevitably poses unprecedented challenges and risks. Among them, AI-generated deceptive content interfering with elections is considered an important challenge faced globally.

On January 23, local time, the Republican primary in New Hampshire for the 2024 U.S. presidential election was held. Before this, many American voters reported receiving a call “from President Biden.”

This call started with Biden’s catchphrase “It’s a bunch of nonsense” and suggested voters not to vote for Trump, but to save their ballots for the November election to vote for the Democrats. Subsequently, White House Press Secretary Pierre clarified that this was a forged phone recording.

Analysts worry that in an era when American voters are easily influenced by misinformation, AI may create more chaos during elections.

According to incomplete statistics, over 70 countries or regions will hold important elections in 2024, covering more than half of the world’s population. At the recently concluded 60th Munich Security Conference, several global tech companies signed agreements to commit to combating AI misuse aimed at interfering with elections in 2024.

Ways to Apply AI

Must Fully Comply with Ethical Rules

According to data from the China Academy of Information and Communications Technology, in 2023, the adoption rate of generative AI among enterprises in China has reached 15%, with a market scale of approximately 14.4 trillion yuan. The adoption rate of generative AI technology has also achieved rapid growth in the four major industries: manufacturing, retail, telecommunications, and healthcare.

Experts predict that by 2035, generative AI is expected to contribute nearly 90 trillion yuan in economic value globally, with China surpassing 30 trillion yuan, accounting for over 40%.

How Should We View the Future Development of Artificial Intelligence?

Li Xiaodong, Vice Chairman of the China Internet Association and founder of Fuxi Think Tank, analyzed that artificial intelligence has gone through 60 to 70 years of development, and is now widely applied in technological innovation, cultural industries, and industrial manufacturing. The enhancement of computing power and reduction of costs have also brought general artificial intelligence closer to ordinary people.

It can be foreseen that in the near future, artificial intelligence will be ubiquitous, driving the transition of information skills from digitalization and networking to a fully intelligent era. “Soon we will no longer discuss artificial intelligence because it will be integrated into our lives and will be everywhere,” said Li Xiaodong.

In a sense, the utilization of artificial intelligence will create new gaps and digital divides between countries, institutions, and even individuals, and drive humanity from agricultural civilization and industrial civilization to digital civilization. Therefore,whether we can fully learn from and utilize artificial intelligence will have a differentiating effect on humanity and may even have a huge impact on human civilization.

What Challenges Will Regulation Face as AI Rapidly Develops?

Li Xiaodong stated, “Data acquisition and application methods” are the two major issues in AI regulation. Reasonable and legal data acquisition is crucial for artificial intelligence, and the application methods of artificial intelligence must also fully comply with ethical rules. If these two core issues are not handled properly, they will severely impact the development and utilization of artificial intelligence.

From the perspective of data acquisition, collection and acquisition not only involve data ownership issues but also relate to national security and personal privacy. How to acquire data reasonably and legally is crucial for artificial intelligence.

In addition, how to effectively connect data faults, promote data exchange and sharing, and enhance the interoperability of data is also a key focus of AI governance. Otherwise, the development of artificial intelligence without continuous data support will be severely impaired.

From the perspective of AI application methods, artificial intelligence demonstrates its powerful information processing capabilities in unprecedented ways, fundamentally enhancing human efficiency and effectiveness in utilizing information. However, human society is bound by specific national and cultural laws and moral constraints, and the development of artificial intelligence must fully comply with these laws and ethics.

Currently, some AI technologies do indeed impact traditional morals and established laws, generating new global ethical norms and rules. During the process of forming these rules, it is essential to maintain proactive interaction and tracking, promoting ethical norms and global rules to advance positively.

Authentic Kambach Non-Coated Titanium Non-Stick Pan (Official Manufacturer Direct Shipping)Jianjian Direct Life Hall559Purchase

Selected Previous Issues:

ChatGPT Special Edition

Learn in 1 Minute What ChatGPT Is

How to Play with the AI Chatbot | c-h-a-t-G-P-T

What is the real challenge of ChatGPT? What impact does it have on us? A few philosophical reflections

What is ChatGPT? The inventor explains it personally

ChatGPT Q&A | Summary of Teacher Work Reports from Various Subjects

ChatGPT Q&A | Please write a work report for Teacher XX

ChatGPT Q&A | My wife says 1+1=8 (See how Microsoft AICopilot responds)

ChatGPT Q&A | What are the strongest AI listed companies in China in terms of algorithms, computing power, and data? (See how Microsoft AI responds)

ChatGPT Q&A | What are the strongest AI listed companies in China in terms of algorithms, computing power, and data?

ChatGPT Q&A | Where to find teaching materials?

ChatGPT Q&A | How many jobs will be replaced by AI?

ChatGPT Q&A | I am 45 years old this year and still haven’t found a job. What should I do?

ChatGPT Q&A | The fridge has half a pound of chicken and mushrooms, introduce related recipes?

ChatGPT Q&A | How do you think ordinary people can have a happy life?